Substitution of a Single Amino Acid Switches the Tentoxin-resistant Thermophilic F1-ATPase into a Tentoxin-sensitive Enzyme*

In contrast to the homologous bacterial and mitochondrial enzymes the chloroplast F1-ATPase (CF1) is strongly affected by the phytopathogenic inhibitor tentoxin. Based on structural information obtained from crystals of a CF1-tentoxin co-complex (Groth, G. (2002) Proc. Natl. Acad. Sci. U. S. A. 99, 3464–3468) we have replaced residues βSer66 and αArg132 in the α3β3γ subcomplex of the thermophilic F1-ATPase from Bacillus PS3 by the corresponding residues of the chloroplast ATPase to confer tentoxin sensitivity to the thermophilic enzyme. The mutation αArg132 → Pro, proposed to relieve steric constraints on tentoxin binding, did not have any significant effect. However, mutation βSer66 → Ala, predicted to provide a crucial hydrogen bond with the inhibitor, resulted in tentoxin inhibition of ATP hydrolysis comparable with the situation found with the chloroplast enzyme.

Tentoxin is a cyclic tetrapeptide derived from phytopathogenic fungi of the Alternaria species, causing chlorosis in sensitive plant species. It acts as an inhibitor of the chloroplast F 0 F 1 -ATP synthase from these species, but not of the homologous enzymes from other bacteria and animals (1)(2)(3)(4). In membrane-bound F 0 F 1 -ATP synthase, both ATP synthesis and ATP hydrolysis are inhibited by tentoxin (1), with the soluble F 1 subcomplex, which is not capable of ATP synthesis; ATP hydrolysis is inhibited (2). Although binding studies suggested an uncompetetive manner of inhibition by interference with cooperative release of nucleotides from the enzyme (2,5), the pre-cise mechanism of tentoxin is not known. Based on labeling studies, one high affinity inhibitory binding site and additionally one to two low affinity binding sites have been proposed (6,7). Binding of tentoxin to low affinity sites relieves inhibition caused by binding to the high affinity site (6,7).
The F 1 subcomplex of F 0 F 1 -ATP synthase is also referred to as F 1 -ATPase and consists of the subunits ␣ 3 ␤ 3 ␥␦⑀. Its ␣ 3 ␤ 3 ␥ subunits make up the smallest entity capable of continuous ATP hydrolysis (8,9). High resolution structures of the ␣ 3 ␤ 3 ␥ complex from bovine heart mitochondria as well as of the ␣ 3 ␤ 3 region from the thermophilic Bacillus PS3 and from spinach chloroplast revealed an alternating, hexagonal arrangement of the three ␣ and three ␤ subunits (10 -12). These subunits consist of three domains: N-terminal ␤-barrels, a central nucleotidebinding domain, and a C-terminal bundle of ␣-helices (10).
Recent results obtained by co-crystallization of spinach chloroplast F 1 -ATPase (CF 1 ) 1 and tentoxin shed more light on the binding of the inhibitor (13) and showed that tentoxin is bound at the ␣␤-interface in a cleft near the N-terminal ␤-barrel domains. The structure of the CF 1 -tentoxin complex suggests a critical role of residue ␤Asp 83 for tentoxin binding and/or inhibition, which has been concluded from mutagenesis experiments in the past (14,15), but it displayed at the same time structural differences in the vicinity of ␤Asp 83 between CF 1 and tentoxin-resistant F 1 -ATPases, e.g. from Escherichia coli (EF 1 ) or from the thermophilic Bacillus PS3 (TF 1 ).
Another critical region for tentoxin inhibition seems to be located in the chloroplast ATPase ␣ subunit. Studies using chimeric ␣ 3 ␤ 3 ␥ complexes that have been assembled from subunits originating from the tentoxin-sensitive CF 1 and from the insensitive Rhodospirillum rubrum F 1 -ATPase (16,17) indicated that the poorly conserved residues ␣120 -133 might be crucial.
In this report we superimposed the structures of the CF 1tentoxin complex (13) and the corresponding parts of TF 1 (11) to pinpoint crucial amino acid residues involved in tentoxin binding. We predicted that ␤Ser 66 and ␣Arg 132 of TF 1 (corresponding to ␤Ala 81 and ␣Pro 133 of CF 1 ) play a central role in conferring tentoxin resistance to TF 1 . To test these predictions based on the static picture provided by the crystal structure, we prepared two mutants of the ␣ 3 ␤ 3 ␥ complex of the tentoxininsensitive TF 1 , where the two critical residues ␤Ser 66 and ␣Arg 132 were replaced by alanine and by proline, respectively, as found in the corresponding position of CF 1 . The results of this mutagenesis study show how inhibition of ATP hydrolysis responds to subtle changes of the protein structure.

EXPERIMENTAL PROCEDURES
Chemicals-Tentoxin was purchased from Sigma, a pyruvate kinase/ lactate dehydrogenase mixture was obtained from Roche Molecular Biochemicals, and restriction enzymes were from New England Biolabs. All other chemicals were of analytical grade.
Plasmid Construction-The pkkHC5 expression plasmid coding for the ␣, ␥, and ␤ subunits of the thermophilic Bacillus PS3 F 1 -ATPase, * This work was supported by the Deutsche Forschungsgemeinschaft (GR1616/4-1). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
** To whom correspondence should be addressed.
carrying a decahistidine tag at the N terminus of the ␤ subunit and a single cysteine in the ␥ subunit (18) (the protein encoded is referred to as "wild-type" TF 1 ␣ 3 ␤ 3 ␥ here) was digested with the restriction enzymes PstI and EcoRV (for the mutation ␤Ser 66 3 Ala) and EcoRI and EcoRV (for the mutation ␣Arg 132 3 Pro). The resulting fragments were then ligated into pBluescript SK vector (Stratagene), previously cut with the same restriction enzymes. Directed mutagenesis was done by the full-circle polymerase chain reaction method with a PCR machine (Tpersonal, Biometra) according to the suggestions from the QuikChange site-directed mutagenesis kit (Stratagene). The primer used were 5Ј-ACAGTACGGACGATCGCCATGGCGGCCACAG ACG-GCCTCATC-3Ј (forward) together with 5Ј-GATGAGGCGTCTGTGGC-CGCCATGGCGATCGTCCGTACTGT-3Ј (backward) for the mutation ␤Ser 66 3 Ala and 5Ј-CGCGCCCGATTGAAAGCCCTGCCCCGGGCGT-TATGGACC-3Ј (forward) with 5Ј-CCGGTCCATAACGCCCGGGGCAG-GGCTTTCAATCGGGCGCG-3Ј (backward) for the mutation ␣Arg 132 3 Pro. The codons carrying the mutation and one base each introducing an additional NcoI (in case of ␤Ser 66 3 Ala) or SmaI (in case of ␣Arg 132 3 Pro) recognition site are underlined. Positive clones were identified by digestion with SmaI or NcoI, respectively, cloned into the equivalent position of the expression vector pkkHC5 and verified by DNA sequencing (ABI Prism 310, PerkinElmer Life Sciences).
Overexpression and Protein Purification-The expression vector was transformed into E. coli JM103⌬uncB-D and cultivated as described in Refs. 9, 18, and 19. Cells were harvested, disrupted by sonification, centrifuged, and the supernatant was subjected to a heat shock for 20 min at 60 o C (9). Proteins were purified with a nickel-nitrilotriacetic acid (Ni-NTA, Quiagen) affinity column (18,19) and stored as an ammonium sulfate precipitate (70% saturation) at 4°C.
ATP Hydrolysis Activity Measurement-ATP hydrolysis activity was determined using an ATP regenerating system (20) with an UV-VIS spectrophotometer (Lambda 40, PerkinElmer Life Sciences). The absorption at 340 nm of a reaction mixture containing 50 mM MOPS/KOH, pH 7.0, 50 mM KCl, 4 mM MgCl 2 , 2 mM phosphoenolpyruvate, 0.2 mM NADH, 2 mM ATP, 40 g/ml pyruvate kinase, 40 g/ml lactate dehydrogenase was measured for 1 min, then the ATP hydrolysis reaction was started by addition of 2 g of ␣ 3 ␤ 3 ␥, and the absorption change was observed for 8 min. Activities were calculated from the slope 2 min after starting the ATP hydrolysis reaction using the extinction coefficient of NADH at 340 nm of 6230 M Ϫ1 cm Ϫ1 .

RESULTS AND DISCUSSION
Superimposition of the CF 1 and TF 1 Structures at the Tentoxin Binding Site-A superimposition of the structure of the CF 1 -tentoxin complex (13) with the structure of the ␣ 3 ␤ 3 subcomplex of TF 1 (11), which is shown in Fig. 1, B and C, revealed that the positions of residues ␣Leu 65 , ␣Va l75 , and ␣Leu 238 , which probably form important hydrophobic contacts with the inhibitor, are essentially conserved as well as the position of the crucial residue Asp 83 in the ␤ subunit (Fig. 1B). Calculation of potential hydrogen bonds showed that the carboxyl side chain of ␤Asp 83 is hydrogen-bonded to the amide hydrogens of leucine 2 and glycine 4 in the tentoxin molecule, which aligns the inhibitor in the binding cleft formed at the ␣␤-interface (13). In TF 1 this critical interaction is probably impaired by a potential hydrogen bond formed between ␤Asp 68 (␤Asp 83 of CF 1 ) and the hydroxyl group of ␤Ser 66 (corresponding to ␤Ala 81 in CF 1 ), which prevents correct tentoxin binding (13). In addition the superimposition of the two F 1 structures clearly visualizes the potential critical role of residue ␣Pro 133 in tentoxin binding in CF 1 . In TF 1 this residue is replaced by arginine (␣Arg 132 ), whose bulky side chain seems to block access to the tentoxin binding site (Fig. 1C).
In contrast, only a slight decrease of activity, amounting to about 10% inhibition, was observed when 20 -100 M tentoxin were added to wild-type TF 1 ␣ 3 ␤ 3 ␥ or the mutant TF 1 ␣ 3 ␤ 3 ␥ (␣R132P) (Fig. 2).  (12,13). Subunit ␣ is colored in yellow, and subunit ␤ is shown in green. B and C, stereo images of the tentoxin binding pocket of CF 1 (13), complexed with one molecule tentoxin (backbone and residue numbers in blue), superimposed with the corresponding part of F 1 -ATPase from the thermophilic Bacillus PS3 (TF 1 , backbone and residue numbers in red) (11). The positions of the residues mutated in this study, ␤Ser 66 (B) and ␣Arg 132 (C), are indicated.
The degree of inhibition, about 70% determined for TF 1 ␣ 3 ␤ 3 ␥ (␤S66A), is comparable with values reported earlier for the Mg-ATPase activity of chloroplast F 1 -ATPase (16,17) and for chimeric mutants constructed by reconstitution of mutated ␣ subunits derived from R. rubrum and ␤ and ␥ subunits derived from CF 1 (16). The K I value measured here was significantly lower than the K I value of about 10 Ϫ8 M published for CF 1 (6, 7), but comparable with values reported for the abovementioned chimeric enzymes (16). A re-activation of the enzyme as observed here in the presence of tentoxin concentrations Ͼ100 M was previously also reported for CF 1 and explained by the binding of a second and possibly a third tentoxin molecule to the F 1 complex, which by an unknown mechanism may relieve inhibition.
The Role of ␤Asp 83 for the Binding of Tentoxin-Functional binding of tentoxin seems to depend essentially on correct hydrogen bonding between the amide hydrogens from the tentoxin backbone and residue ␤Asp 83 (13). In the thermophilic F 1 this important hydrogen bonding is obviously affected by a potential hydrogen bond formed between ␤Asp 68 and the side chain of the adjacent residue ␤Ser 66 (3.3 Å). A similar competition for intermolecular (␤Asp-TTX) and intramolecular hydrogen bonding (␤66 -68) is avoided in the tentoxin-sensitive CF 1 complex as the chloroplast ␤ subunit contains alanine in the equivalent position of the binding site (␤Ala 81 ). For the same reason tentoxin sensitivity can probably be achieved in the F 1 complex from Chlamydomonas reinhardii simply by the replacement of ␤Glu 83 by aspartate (15), as a proline residue, which has no capability to form hydrogen bonds, is located in the adjacent (n-2) position.
A steric effect on the binding of the inhibitor caused by the side chain located in position 81 seems unlikely as alanine and serine show about the same surface volume of 89 Å 3 . In addition the even more bulky threonine (surface volume 116 Å 3 ) or proline side chain (surface volume 113 Å 3 ) is found in the equivalent position of the binding side in the tentoxin-sensitive F 1 complex from Syneccococcus PC6301 or in the Chlamydomonas ␤Glu 83 3 Asp F 1 mutant (15). Thus the structural requirement in the ␤ subunit for effective tentoxin binding is apparently to avoid any intramolecular hydrogen bonding with the crucial aspartate in position 83.
Steric Blockage of the Tentoxin Binding Site Caused by ␣Arg 132 -Although the available structural information ( Fig.  1C; see also Ref. 13) strongly suggested that in wild-type TF 1 ␣ 3 ␤ 3 ␥ the bulky side chain ␣Arg 132 blocks access to the tentoxin binding niche; its replacement by proline did not have a signif-icant effect on tentoxin sensitivity. Furthermore, the double mutant TF 1 ␣ 3 ␤ 3 ␥ (␣R132P/␤S66A) displayed the same tentoxin sensitivity as the single mutant TF 1 ␣ 3 ␤ 3 ␥ (␤S66A) (data not shown), indicating that steric hindrance by this bulky side chain is not a predominant factor for tentoxin binding. The reason why the mutation failed to show the result expected from the CF 1 and TF 1 structures might be related to a substitution of ␣Leu 125 by proline in the native TF 1 complex (surface volume: Leu, 167 Å 3 ; Pro, 112 Å 3 ), which might compensate for the effect caused by the more bulky arginine side chain in position 132. In addition, the wild-type-like activity of the mutant might be explained by dynamic movements of this part of the ␣ subunit during tentoxin binding, which are not visible in a static protein structure, but might be resolved by a set of intermediate structures (alternative conformations) or by dynamic studies.
Requirements for Tentoxin Binding in F 1 -The point mutation ␤Ser 66 3 Ala is sufficient to achieve maximal inhibition, underscoring the importance of the ability of the crucial residue ␤Asp 68 (␤Asp 83 in CF 1 ) to form hydrogen bonds with the tentoxin peptide backbone. The results from our mutagenesis studies indicate that prerequisites for inhibition by tentoxin are a tentoxin binding cleft, an aspartate side chain for correct hydrogen binding, and the absence of other residues that might interfere with this crucial hydrogen bond. The comparably high K I value determined here for tentoxin inhibition indicates that other amino acid residues, which probably are located in the ␣ subunit, also influence the affinity for tentoxin. Experiments to elucidate the role of these residues in fine-tuning the affinity for tentoxin are presently under way in our laboratory.
The results presented in this paper demonstrate in a remarkable way the feasibility of functional predictions based on structural information, e.g. for the design of special characteristics in a target protein. On the other hand, they also stress that, as it comes to functional considerations, structural dynamics should be taken into account.